Efficient Re-Clustering for Convolutional Neural Networks

DOI

10.2991/ceis-16.2016.22How to use a DOI?

Keywords

face recognition; t-distribution; convolutional network; re-clustering; dimensional reduction

Abstract

With the rapid development of face recognition, it has been widely applied to various fields and scenes. Due to the unstable light condition, rotation and occlusion, the stability of the recognition performance is still an issue. In this work, we study the VGG model and analyze features' distribution extracted with the VGG model. To achieving good performance, we suggest a t-distribution based VGG model to reduce the dimension of features and re-clustering them. Furthermore, by recording the path of dimension reduction in training phase, the computational complexity of the test phase is decreased. The proposed algorithm is evaluated on public available datasets. The experimental results demonstrate the significant improvement of the proposed approach on both the simplification of feature space and the efficiency of recognition, especially on limited data.

Copyright

© 2017, the Authors. Published by Atlantis Press.

Open Access

This is an open access article distributed under the CC BY-NC license (http://creativecommons.org/licenses/by-nc/4.0/).

Download article (PDF)

TY  - CONF
AU  - Cheng-Ying Wang
AU  - Feng Cen
PY  - 2016/11
DA  - 2016/11
TI  - Efficient Re-Clustering for Convolutional Neural Networks
BT  - Proceedings of the 2016 International Conference on Computer Engineering and Information Systems
PB  - Atlantis Press
SP  - 111
EP  - 116
SN  - 2352-538X
UR  - https://doi.org/10.2991/ceis-16.2016.22
DO  - 10.2991/ceis-16.2016.22
ID  - Wang2016/11
ER  -